mcp-upbank
Enables querying Up Bank account transactions and spending habits through natural language, using the Up Bank API in read-only mode.
README
MCP-Upbank
A Model Context Protocol server implementation for accessing the Up Bank API
The goal is to test the Up Bank API to see if we can find some use for it in daily LLM usage. Some ideas:
- Get a list of transactions for the day and add to your daily journaling
- Compare your spending habits across months
- Get Navishkar (the author of this repo) a job so he can make the kind of money that allows him to finally afford his own home in Melbourne
Note: The API does not allow mutations so there's no risk of you accidentally deleting your bank account or sending money to a Nigerian prince.
How it works.
Follow Up Bank's own Getting Started guide. You'll want to create a Personal Access Token.
MCP Server's remain a pain to install. I'm going to link to each client's instructions (or the ones I use anyway)
The easiest is via Smithery.ai which gives you the exact config to include in your client.
[!CAUTION]
YourPersonal Access Tokenis Personal. Do not share with others, including adding it inside the Smithery interface. Just paste some random values to see how the config will look.
How its built
I use the openapi spec found on Github to generate zod schemas and types using Orval.
Unfortunately, the api uses opaque cursors which means I can't simply use the outputs generated from Orval.
This is why, for the moment, I'm manually copying some files across and generating a pagination handler myself.
What's next
- Fix/Enhance Orval's generator to make this a better experience. Or maybe I just need to read their docs better.
- Create prompts that provide instructions on analyzing ingested data.
- Distribute executables with Homebrew, Chocolatey, apt etc.
- Resolve the additional endpoints that aren't needed e.g. patch, post, delete on resources
Resources:
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。